Mature human language technology is currently available for only a few of the approximately 7,000 languages in the world. This project aims to develop technology for low-resource languages by projecting information from related languages. Projections are made by drawing on rich networks expressing language similarity information and applying large-scale graph-based machine learning techniques.